Subsection5.4.1Evaluating Limits of Indeterminate Forms
Taylor series provide another method of evaluating limits of indeterminate forms like \(\frac{0}{0}\) or \(\frac{\infty}{\infty}\text{.}\) Previously, to do this, you learned:
However, there are some situations where LβHopitalβs rule is complicated, because the derivatives involved are complicated. Taylor series give a more flexible, general, and simple method.
Sometimes it is not obvious how many terms of the Taylor series to use. A good rule of thumb is 3 or 4 terms, but when in doubt, you can always include more terms. The dots (\(\dots\)) stand for powers of \(x\) greater than the last power that appears.
Up to some decimal places, like accurate up to 3 decimal places, or 5 decimal places. Note that accurate up to \(k\) decimal places basically means to keep the error below \(0.5 \times 10^{-k}\text{.}\) For example:
Add terms one at a time until your calculator output is stable. Add the first term, then the 2nd, then the 3rd, and so on, one at a time. Each time, check if the decimal digits you care about stop changing. After they stay the same for 2 or 3 steps in a row, that is your answer.
This is the most naive and simple method, which will give you the correct answer for most exam problems. However, it is technically not rigorous, and gives misleading results in some rare tricky cases. To be more safe, you can add more terms to confirm that your answer is correct.
Alternating series error bound. If the series is alternating (which a lot of them are), then the error is bounded by the first omitted term. In other words, \(\abs{r_n} \leq a_{n+1}\text{.}\)
If you want error less than \(10^{-4}\text{,}\) then test values of \(n\) such that \(a_{n+1}\) is less than \(10^{-4}\text{.}\)